Installation

Note

LOTUS requires Python 3.7.*

This code uses Theano as backend tool to realize MCMC sampling via PyMC3 but Theano has been deprecated after version 1.0.5 and has been transferred into Aesara for PyMC3>=3.10.0. In our future version, we will update this code with up-to-date PyMC3 but now we choose to use previous version, which need to freeze python==3.7. Therefore before installing this code, it is recommended that create an independent environment via anaconda:

conda create -n lotus python=3.7

and activate it:

conda activate lotus

Using Pip

The recommended method of installing LOTUS is with pip:

python -m pip install lotus-nlte==0.1.1

From Source

LOTUS can be downloaded and installed from GitHub source code by running:

git clone https://github.com/Li-Yangyang/LOTUS
cd LOTUS
python -m pip install -e .

The following dependencies are required to install it successfully:

• cython ==0.29.24

• numpy>=1.16.4

• pandas ==0.24.2

• tables ==3.7.0

• scipy>=1.5.0

• scikit-learn ==0.23.2

• sympy>=1.6.2

• pymc3 ==3.7

• theano ==1.0.4

• astropy>=3.2.1

• h5py>=2.10.0

• joblib ==1.0.1

• numdifftools ==0.9.39

• matplotlib>=3.1.3 (for plotting)

• tqdm

• corner

• requests

The rest will be installed as well but are used for future more ways of interpolation:

• torch

• gpytorch

• rbf

You can install them by:

python -m pip install "lotus-nlte[advanced-interp]"